Dear Perpass WG,

Several have expressed valid concerns regarding abuse of encrypted email.  
While CAs may not be trustworthy, it seems DANE offers a workable and 
reasonably transparent  alternative to protect against undetected tampering.  
In addition, DANE reduces incremental costs associated with certificates which 
should prove beneficial for email use.

That said, proprietary schemes are available permitting multiple decoding keys 
derived from conveyed indexes.  The derived keys are assigned to outbound 
message scanners used by enterprises to ensure governmental data compliance 
requirements while also ensuring the integrity of the entire path traversed.  
If there is interest, I could prevail on my employer to disclose IPR terms and 
details.

Secondly, while web based TLS exchanges normally ignore client certificates, 
this would not be desirable for StartTLS related to open email exchanges.  
Currently, only source IP addresses are effectively used to defend SMTP 
servers.  DKIM does not offer a suitable replacement because it fails to 
capture who initiated the exchange, and to whom it was being sent.  In other 
words, DKIM lacks essential elements needed to properly identify those 
accountable for email abuse.

IPv6 will significantly challenge the use of source IP addresses.  While the 
number of legitimate addresses may represent a reasonable number, most of email 
is sourced from compromised systems likely using privacy extensions.  Just a 
bit map tracking whether a /64 prefix is active requires 5,650 Terra-bytes to 
cover just the announced /64 prefix space.  At any point in time, a bit more 
than 100 million registered domains are active that collapses down to 
subdomains below the registrar.  This relatively small number is fairly 
manageable compared against IP addresses, even when attempting to just 
white-list new MTAs.

A domain based approach may seem fairly disruptive, but even the best content 
scanners fail to provide full detections while demanding significant resources. 
Content based acceptance is not cost effective as the first stage in a vetting 
process.  One hundred thousand domains control 90% of the Internet traffic. The 
top 150 domains control 50%, and the top 2,500 domains control 75%. 

Secure SMTP using DNS-Based Authentication of Named Entities (DANE) TLSA 
records and SMTP security via opportunistic DANE TLS offer interesting starting 
points.  For this to work well, a more disruptive approach is required where 
sending domains should be encouraged to use their own certificates.  The 
initial availability of TSLA RRs should not miss the opportunity to use this to 
signal a new paradigm of expectations.

Regards.
Douglas Otis

_______________________________________________
perpass mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/perpass

Reply via email to